A flexible rank-based framework for detecting copy number aberrations from array data

Motivation: DNA copy number aberration-both inherited and sporadic-is a significant contributor to a variety of human diseases. Copy number characterization is therefore an area of intense research. Probe hybridization-based arrays are important tools used to measure copy number in a high-throughput manner. Results: In this article, we present a simple but powerful nonparametric rank-based approach to detect deletions and gains from raw array copy number measurements. We use three different rank-based statistics to detect three separate molecular phenomena-somatic lesions, germline deletions and germline gains. The approach is robust and rigorously grounded in statistical theory, thereby enabling the meaningful assignment of statistical significance to each putative aberration. We demonstrate the flexibility of our approach by applying it to data from three different array platforms. We show that our method compares favorably with established approaches by applying it to published well-characterized samples. Power simulations demonstrate exquisite sensitivity for array data of reasonable quality. Conclusions: Our flexible rank-based framework is suitable for multiple platforms including single nucleotide polymorphism arrays and array comparative genomic hybridization, and can reliably detect gains or losses of genomic DNA, whether inherited, de novo, or somatic.